Stored process performance and increasing user load

[samira]

Hello,

I have a question regarding SAS Stored processes perforemance. before asking question, I want to brifely explain about our application in SAS. I am working on a montoring system which I have developed using html and javascript wrote in SAS stored process. first I have coded some analysis (SAS programming) and the result will be sent as json documents to the stored process which contain HTML and JavaScript code. 

Now we have decided to grant access to some users in department.  But the question is how many users would be able to connect to application at the same time without having bottleneck. On the other words how much sever can tolerate while user load increase?

I am Also wondering if the way that I coded my stored process will affect the perforemance.

Thank you in advance!

[jklaverstijn]

Hi Samira,

How much your server can handle is entirely dependent upon its specifications and difficult for us to answer. Also network speeds and other infra components will determine the user experience.

We use automated testing tools to create a predetermined work load and measure/observer the server peformance and browser response times. There are many of these products available. They take the guess work out of the task and prevent you from having to gather a large group of people to manually create a workload, from simple to complicated/advanced, from free (as in beer) to expensive. JMeter comes to mind.

In order to match the workload you can load balance your STP servers across multiple servers if need be. So you can scale out as well as up. Do keep the terms of your SAS license in mind.

Hope this helps,

- Jan

[samira]

Thank you Jan for your Answer,

So the way that we wrote SAS program does not play an important role on this right? for example the number of datasteps

[jklaverstijn]

But of course it does. The efficiency of your code is paramount. But also if you do not scale it should be a concern. This is a very broad area of interest and many vooks were qrittem on it. Even specific to SAS. As wven very effcient programs can take a long time it is hard to say without being intimate to your infrastructure, data and code. When in doubt invite a set of expert eyes from outside.